1 /* $Id: mdoc_argv.c,v 1.86 2012/11/18 00:05:35 schwarze Exp $ */
3 * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
4 * Copyright (c) 2012 Ingo Schwarze <schwarze@openbsd.org>
6 * Permission to use, copy, modify, and distribute this software for any
7 * purpose with or without fee is hereby granted, provided that the above
8 * copyright notice and this permission notice appear in all copies.
10 * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
11 * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
12 * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
13 * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
14 * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
15 * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
16 * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
22 #include <sys/types.h>
32 #include "libmandoc.h"
34 #define MULTI_STEP 5 /* pre-allocate argument values */
35 #define DELIMSZ 6 /* max possible size of a delimiter */
39 ARGSFL_DELIM, /* handle delimiters of [[::delim::][ ]+]+ */
40 ARGSFL_TABSEP /* handle tab/`Ta' separated phrases */
44 ARGV_NONE, /* no args to flag (e.g., -split) */
45 ARGV_SINGLE, /* one arg to flag (e.g., -file xxx) */
46 ARGV_MULTI /* multiple args (e.g., -column xxx yyy) */
51 const enum mdocargt *argvs;
54 static void argn_free(struct mdoc_arg *, int);
55 static enum margserr args(struct mdoc *, int, int *,
56 char *, enum argsflag, char **);
57 static int args_checkpunct(const char *, int);
58 static int argv_multi(struct mdoc *, int,
59 struct mdoc_argv *, int *, char *);
60 static int argv_single(struct mdoc *, int,
61 struct mdoc_argv *, int *, char *);
63 static const enum argvflag argvflags[MDOC_ARG_MAX] = {
64 ARGV_NONE, /* MDOC_Split */
65 ARGV_NONE, /* MDOC_Nosplit */
66 ARGV_NONE, /* MDOC_Ragged */
67 ARGV_NONE, /* MDOC_Unfilled */
68 ARGV_NONE, /* MDOC_Literal */
69 ARGV_SINGLE, /* MDOC_File */
70 ARGV_SINGLE, /* MDOC_Offset */
71 ARGV_NONE, /* MDOC_Bullet */
72 ARGV_NONE, /* MDOC_Dash */
73 ARGV_NONE, /* MDOC_Hyphen */
74 ARGV_NONE, /* MDOC_Item */
75 ARGV_NONE, /* MDOC_Enum */
76 ARGV_NONE, /* MDOC_Tag */
77 ARGV_NONE, /* MDOC_Diag */
78 ARGV_NONE, /* MDOC_Hang */
79 ARGV_NONE, /* MDOC_Ohang */
80 ARGV_NONE, /* MDOC_Inset */
81 ARGV_MULTI, /* MDOC_Column */
82 ARGV_SINGLE, /* MDOC_Width */
83 ARGV_NONE, /* MDOC_Compact */
84 ARGV_NONE, /* MDOC_Std */
85 ARGV_NONE, /* MDOC_Filled */
86 ARGV_NONE, /* MDOC_Words */
87 ARGV_NONE, /* MDOC_Emphasis */
88 ARGV_NONE, /* MDOC_Symbolic */
89 ARGV_NONE /* MDOC_Symbolic */
92 static const enum mdocargt args_Ex[] = {
97 static const enum mdocargt args_An[] = {
103 static const enum mdocargt args_Bd[] = {
115 static const enum mdocargt args_Bf[] = {
122 static const enum mdocargt args_Bk[] = {
127 static const enum mdocargt args_Bl[] = {
146 static const struct mdocarg mdocargs[MDOC_MAX] = {
147 { ARGSFL_DELIM, NULL }, /* Ap */
148 { ARGSFL_NONE, NULL }, /* Dd */
149 { ARGSFL_NONE, NULL }, /* Dt */
150 { ARGSFL_NONE, NULL }, /* Os */
151 { ARGSFL_NONE, NULL }, /* Sh */
152 { ARGSFL_NONE, NULL }, /* Ss */
153 { ARGSFL_NONE, NULL }, /* Pp */
154 { ARGSFL_DELIM, NULL }, /* D1 */
155 { ARGSFL_DELIM, NULL }, /* Dl */
156 { ARGSFL_NONE, args_Bd }, /* Bd */
157 { ARGSFL_NONE, NULL }, /* Ed */
158 { ARGSFL_NONE, args_Bl }, /* Bl */
159 { ARGSFL_NONE, NULL }, /* El */
160 { ARGSFL_NONE, NULL }, /* It */
161 { ARGSFL_DELIM, NULL }, /* Ad */
162 { ARGSFL_DELIM, args_An }, /* An */
163 { ARGSFL_DELIM, NULL }, /* Ar */
164 { ARGSFL_DELIM, NULL }, /* Cd */
165 { ARGSFL_DELIM, NULL }, /* Cm */
166 { ARGSFL_DELIM, NULL }, /* Dv */
167 { ARGSFL_DELIM, NULL }, /* Er */
168 { ARGSFL_DELIM, NULL }, /* Ev */
169 { ARGSFL_NONE, args_Ex }, /* Ex */
170 { ARGSFL_DELIM, NULL }, /* Fa */
171 { ARGSFL_NONE, NULL }, /* Fd */
172 { ARGSFL_DELIM, NULL }, /* Fl */
173 { ARGSFL_DELIM, NULL }, /* Fn */
174 { ARGSFL_DELIM, NULL }, /* Ft */
175 { ARGSFL_DELIM, NULL }, /* Ic */
176 { ARGSFL_DELIM, NULL }, /* In */
177 { ARGSFL_DELIM, NULL }, /* Li */
178 { ARGSFL_NONE, NULL }, /* Nd */
179 { ARGSFL_DELIM, NULL }, /* Nm */
180 { ARGSFL_DELIM, NULL }, /* Op */
181 { ARGSFL_NONE, NULL }, /* Ot */
182 { ARGSFL_DELIM, NULL }, /* Pa */
183 { ARGSFL_NONE, args_Ex }, /* Rv */
184 { ARGSFL_DELIM, NULL }, /* St */
185 { ARGSFL_DELIM, NULL }, /* Va */
186 { ARGSFL_DELIM, NULL }, /* Vt */
187 { ARGSFL_DELIM, NULL }, /* Xr */
188 { ARGSFL_NONE, NULL }, /* %A */
189 { ARGSFL_NONE, NULL }, /* %B */
190 { ARGSFL_NONE, NULL }, /* %D */
191 { ARGSFL_NONE, NULL }, /* %I */
192 { ARGSFL_NONE, NULL }, /* %J */
193 { ARGSFL_NONE, NULL }, /* %N */
194 { ARGSFL_NONE, NULL }, /* %O */
195 { ARGSFL_NONE, NULL }, /* %P */
196 { ARGSFL_NONE, NULL }, /* %R */
197 { ARGSFL_NONE, NULL }, /* %T */
198 { ARGSFL_NONE, NULL }, /* %V */
199 { ARGSFL_DELIM, NULL }, /* Ac */
200 { ARGSFL_NONE, NULL }, /* Ao */
201 { ARGSFL_DELIM, NULL }, /* Aq */
202 { ARGSFL_DELIM, NULL }, /* At */
203 { ARGSFL_DELIM, NULL }, /* Bc */
204 { ARGSFL_NONE, args_Bf }, /* Bf */
205 { ARGSFL_NONE, NULL }, /* Bo */
206 { ARGSFL_DELIM, NULL }, /* Bq */
207 { ARGSFL_DELIM, NULL }, /* Bsx */
208 { ARGSFL_DELIM, NULL }, /* Bx */
209 { ARGSFL_NONE, NULL }, /* Db */
210 { ARGSFL_DELIM, NULL }, /* Dc */
211 { ARGSFL_NONE, NULL }, /* Do */
212 { ARGSFL_DELIM, NULL }, /* Dq */
213 { ARGSFL_DELIM, NULL }, /* Ec */
214 { ARGSFL_NONE, NULL }, /* Ef */
215 { ARGSFL_DELIM, NULL }, /* Em */
216 { ARGSFL_NONE, NULL }, /* Eo */
217 { ARGSFL_DELIM, NULL }, /* Fx */
218 { ARGSFL_DELIM, NULL }, /* Ms */
219 { ARGSFL_DELIM, NULL }, /* No */
220 { ARGSFL_DELIM, NULL }, /* Ns */
221 { ARGSFL_DELIM, NULL }, /* Nx */
222 { ARGSFL_DELIM, NULL }, /* Ox */
223 { ARGSFL_DELIM, NULL }, /* Pc */
224 { ARGSFL_DELIM, NULL }, /* Pf */
225 { ARGSFL_NONE, NULL }, /* Po */
226 { ARGSFL_DELIM, NULL }, /* Pq */
227 { ARGSFL_DELIM, NULL }, /* Qc */
228 { ARGSFL_DELIM, NULL }, /* Ql */
229 { ARGSFL_NONE, NULL }, /* Qo */
230 { ARGSFL_DELIM, NULL }, /* Qq */
231 { ARGSFL_NONE, NULL }, /* Re */
232 { ARGSFL_NONE, NULL }, /* Rs */
233 { ARGSFL_DELIM, NULL }, /* Sc */
234 { ARGSFL_NONE, NULL }, /* So */
235 { ARGSFL_DELIM, NULL }, /* Sq */
236 { ARGSFL_NONE, NULL }, /* Sm */
237 { ARGSFL_DELIM, NULL }, /* Sx */
238 { ARGSFL_DELIM, NULL }, /* Sy */
239 { ARGSFL_DELIM, NULL }, /* Tn */
240 { ARGSFL_DELIM, NULL }, /* Ux */
241 { ARGSFL_DELIM, NULL }, /* Xc */
242 { ARGSFL_NONE, NULL }, /* Xo */
243 { ARGSFL_NONE, NULL }, /* Fo */
244 { ARGSFL_DELIM, NULL }, /* Fc */
245 { ARGSFL_NONE, NULL }, /* Oo */
246 { ARGSFL_DELIM, NULL }, /* Oc */
247 { ARGSFL_NONE, args_Bk }, /* Bk */
248 { ARGSFL_NONE, NULL }, /* Ek */
249 { ARGSFL_NONE, NULL }, /* Bt */
250 { ARGSFL_NONE, NULL }, /* Hf */
251 { ARGSFL_NONE, NULL }, /* Fr */
252 { ARGSFL_NONE, NULL }, /* Ud */
253 { ARGSFL_DELIM, NULL }, /* Lb */
254 { ARGSFL_NONE, NULL }, /* Lp */
255 { ARGSFL_DELIM, NULL }, /* Lk */
256 { ARGSFL_DELIM, NULL }, /* Mt */
257 { ARGSFL_DELIM, NULL }, /* Brq */
258 { ARGSFL_NONE, NULL }, /* Bro */
259 { ARGSFL_DELIM, NULL }, /* Brc */
260 { ARGSFL_NONE, NULL }, /* %C */
261 { ARGSFL_NONE, NULL }, /* Es */
262 { ARGSFL_NONE, NULL }, /* En */
263 { ARGSFL_DELIM, NULL }, /* Dx */
264 { ARGSFL_NONE, NULL }, /* %Q */
265 { ARGSFL_NONE, NULL }, /* br */
266 { ARGSFL_NONE, NULL }, /* sp */
267 { ARGSFL_NONE, NULL }, /* %U */
268 { ARGSFL_NONE, NULL }, /* Ta */
273 * Parse an argument from line text. This comes in the form of -key
274 * [value0...], which may either have a single mandatory value, at least
275 * one mandatory value, an optional single value, or no value.
278 mdoc_argv(struct mdoc *mdoc, int line, enum mdoct tok,
279 struct mdoc_arg **v, int *pos, char *buf)
282 struct mdoc_argv tmp;
283 struct mdoc_arg *arg;
284 const enum mdocargt *ap;
286 if ('\0' == buf[*pos])
288 else if (NULL == (ap = mdocargs[tok].argvs))
290 else if ('-' != buf[*pos])
293 /* Seek to the first unescaped space. */
299 for ( ; buf[*pos] ; (*pos)++)
300 if (' ' == buf[*pos] && '\\' != buf[*pos - 1])
304 * We want to nil-terminate the word to look it up (it's easier
305 * that way). But we may not have a flag, in which case we need
306 * to restore the line as-is. So keep around the stray byte,
307 * which we'll reset upon exiting (if necessary).
310 if ('\0' != (sv = buf[*pos]))
311 buf[(*pos)++] = '\0';
314 * Now look up the word as a flag. Use temporary storage that
315 * we'll copy into the node's flags, if necessary.
318 memset(&tmp, 0, sizeof(struct mdoc_argv));
322 tmp.arg = MDOC_ARG_MAX;
324 while (MDOC_ARG_MAX != (tmp.arg = *ap++))
325 if (0 == strcmp(p, mdoc_argnames[tmp.arg]))
328 if (MDOC_ARG_MAX == tmp.arg) {
330 * The flag was not found.
331 * Restore saved zeroed byte and return as a word.
338 /* Read to the next word (the argument). */
340 while (buf[*pos] && ' ' == buf[*pos])
343 switch (argvflags[tmp.arg]) {
345 if ( ! argv_single(mdoc, line, &tmp, pos, buf))
349 if ( ! argv_multi(mdoc, line, &tmp, pos, buf))
356 if (NULL == (arg = *v))
357 arg = *v = mandoc_calloc(1, sizeof(struct mdoc_arg));
360 arg->argv = mandoc_realloc
361 (arg->argv, arg->argc * sizeof(struct mdoc_argv));
363 memcpy(&arg->argv[(int)arg->argc - 1],
364 &tmp, sizeof(struct mdoc_argv));
370 mdoc_argv_free(struct mdoc_arg *p)
384 for (i = (int)p->argc - 1; i >= 0; i--)
392 argn_free(struct mdoc_arg *p, int iarg)
394 struct mdoc_argv *arg;
397 arg = &p->argv[iarg];
399 if (arg->sz && arg->value) {
400 for (j = (int)arg->sz - 1; j >= 0; j--)
405 for (--p->argc; iarg < (int)p->argc; iarg++)
406 p->argv[iarg] = p->argv[iarg+1];
410 mdoc_zargs(struct mdoc *mdoc, int line, int *pos, char *buf, char **v)
413 return(args(mdoc, line, pos, buf, ARGSFL_NONE, v));
417 mdoc_args(struct mdoc *mdoc, int line, int *pos,
418 char *buf, enum mdoct tok, char **v)
423 fl = mdocargs[tok].flags;
426 return(args(mdoc, line, pos, buf, fl, v));
429 * We know that we're in an `It', so it's reasonable to expect
430 * us to be sitting in a `Bl'. Someday this may not be the case
431 * (if we allow random `It's sitting out there), so provide a
432 * safe fall-back into the default behaviour.
435 for (n = mdoc->last; n; n = n->parent)
436 if (MDOC_Bl == n->tok)
437 if (LIST_column == n->norm->Bl.type) {
442 return(args(mdoc, line, pos, buf, fl, v));
446 args(struct mdoc *mdoc, int line, int *pos,
447 char *buf, enum argsflag fl, char **v)
452 if ('\0' == buf[*pos]) {
453 if (MDOC_PPHRASE & mdoc->flags)
456 * If we're not in a partial phrase and the flag for
457 * being a phrase literal is still set, the punctuation
460 if (MDOC_PHRASELIT & mdoc->flags)
461 mdoc_pmsg(mdoc, line, *pos, MANDOCERR_BADQUOTE);
463 mdoc->flags &= ~MDOC_PHRASELIT;
469 if (ARGSFL_DELIM == fl)
470 if (args_checkpunct(buf, *pos))
474 * First handle TABSEP items, restricted to `Bl -column'. This
475 * ignores conventional token parsing and instead uses tabs or
476 * `Ta' macros to separate phrases. Phrases are parsed again
477 * for arguments at a later phase.
480 if (ARGSFL_TABSEP == fl) {
481 /* Scan ahead to tab (can't be escaped). */
482 p = strchr(*v, '\t');
485 /* Scan ahead to unescaped `Ta'. */
486 if ( ! (MDOC_PHRASELIT & mdoc->flags))
487 for (pp = *v; ; pp++) {
488 if (NULL == (pp = strstr(pp, "Ta")))
490 if (pp > *v && ' ' != *(pp - 1))
492 if (' ' == *(pp + 2) || '\0' == *(pp + 2))
496 /* By default, assume a phrase. */
500 * Adjust new-buffer position to be beyond delimiter
501 * mark (e.g., Ta -> end + 2).
504 *pos += pp < p ? 2 : 1;
505 rc = pp < p ? ARGS_PHRASE : ARGS_PPHRASE;
507 } else if (p && ! pp) {
510 } else if (pp && ! p) {
518 /* Whitespace check for eoln case... */
519 if ('\0' == *p && ' ' == *(p - 1))
520 mdoc_pmsg(mdoc, line, *pos, MANDOCERR_EOLNSPACE);
522 *pos += (int)(p - *v);
524 /* Strip delimiter's preceding whitespace. */
526 while (pp > *v && ' ' == *pp) {
527 if (pp > *v && '\\' == *(pp - 1))
533 /* Strip delimiter's proceeding whitespace. */
534 for (pp = &buf[*pos]; ' ' == *pp; pp++, (*pos)++)
541 * Process a quoted literal. A quote begins with a double-quote
542 * and ends with a double-quote NOT preceded by a double-quote.
543 * Whitespace is NOT involved in literal termination.
546 if (MDOC_PHRASELIT & mdoc->flags || '\"' == buf[*pos]) {
547 if ( ! (MDOC_PHRASELIT & mdoc->flags))
550 if (MDOC_PPHRASE & mdoc->flags)
551 mdoc->flags |= MDOC_PHRASELIT;
553 for ( ; buf[*pos]; (*pos)++) {
554 if ('\"' != buf[*pos])
556 if ('\"' != buf[*pos + 1])
561 if ('\0' == buf[*pos]) {
562 if (MDOC_PPHRASE & mdoc->flags)
564 mdoc_pmsg(mdoc, line, *pos, MANDOCERR_BADQUOTE);
568 mdoc->flags &= ~MDOC_PHRASELIT;
569 buf[(*pos)++] = '\0';
571 if ('\0' == buf[*pos])
574 while (' ' == buf[*pos])
577 if ('\0' == buf[*pos])
578 mdoc_pmsg(mdoc, line, *pos, MANDOCERR_EOLNSPACE);
584 *v = mandoc_getarg(mdoc->parse, &p, line, pos);
590 * Check if the string consists only of space-separated closing
591 * delimiters. This is a bit of a dance: the first must be a close
592 * delimiter, but it may be followed by middle delimiters. Arbitrary
593 * whitespace may separate these tokens.
596 args_checkpunct(const char *buf, int i)
602 /* First token must be a close-delimiter. */
604 for (j = 0; buf[i] && ' ' != buf[i] && j < DELIMSZ; j++, i++)
611 if (DELIM_CLOSE != mdoc_isdelim(dbuf))
614 while (' ' == buf[i])
617 /* Remaining must NOT be open/none. */
621 while (buf[i] && ' ' != buf[i] && j < DELIMSZ)
622 dbuf[j++] = buf[i++];
628 d = mdoc_isdelim(dbuf);
629 if (DELIM_NONE == d || DELIM_OPEN == d)
632 while (' ' == buf[i])
636 return('\0' == buf[i]);
640 argv_multi(struct mdoc *mdoc, int line,
641 struct mdoc_argv *v, int *pos, char *buf)
646 for (v->sz = 0; ; v->sz++) {
647 if ('-' == buf[*pos])
649 ac = args(mdoc, line, pos, buf, ARGSFL_NONE, &p);
650 if (ARGS_ERROR == ac)
652 else if (ARGS_EOLN == ac)
655 if (0 == v->sz % MULTI_STEP)
656 v->value = mandoc_realloc(v->value,
657 (v->sz + MULTI_STEP) * sizeof(char *));
659 v->value[(int)v->sz] = mandoc_strdup(p);
666 argv_single(struct mdoc *mdoc, int line,
667 struct mdoc_argv *v, int *pos, char *buf)
672 ac = args(mdoc, line, pos, buf, ARGSFL_NONE, &p);
673 if (ARGS_ERROR == ac)
679 v->value = mandoc_malloc(sizeof(char *));
680 v->value[0] = mandoc_strdup(p);